Industrial devices are periodically examined to maintain safety. Technologies for monitoring the state of devices by acceleration sensors or temperature sensors have been put into practical use to promptly detect abnormality in operation and prevent an accident. In many cases, mechanical abnormality is noticed by noise. Thus, as a method of monitoring the state of devices based on operating noise, technologies for analyzing acoustic data and detecting a mechanical failure have been developed.
When the device to be monitored includes a component which produces an airflow, such as a fan, a microphone may be exposed to the airflow. Thus, the signal of air noise may be mixed in the signal of sound collected by the microphone. Therefore, a new technology for accurately detecting the abnormality of the device by reducing the influence of air noise is required.